Our study found that the occurrence of pendelluft during T piece spontaneous breathing happened in 70.4% ICU patients who were difficult to wean. The patients with pendelluft had significantly longer mechanical ventilation duration, shorter VFDs to day 28 and higher 28-day mortality compared with patients without pendelluft. Patients who were successfully extubated exhibited a significantly lower occurrence of pendelluft compared to those who were not. Our findings further revealed that the occurrence of pendelluft was independently associated with an increased risk of 28-day mortality among these patients.

Our previous study revealed a pendelluft prevalence of 31% in a population with acute respiratory failure under mechanical ventilation [14]. However, it’s important to note that EIT assessments were not conducted during SBT in that study. In our current study of difficult-to-wean population, we observed a notable increase in the prevalence of pendelluft, rising to 70.4% during SBT. The occurrence of the pendelluft may be attributed to factors such as alveolar heterogeneity or excessive spontaneous breathing effort, leading to gas movement within the lung [7, 8, 21, 22]. Throughout the weaning process, patients’ spontaneous breathing effort progressively intensifies, reaching its zenith when they are disconnected from the ventilator. During spontaneous breathing with a T piece, the local negative pleural pressure generated by diaphragmatic muscle contraction can result in inflation of the dependent lung region and simultaneous deflation of the nondependent lung region [23]. This dynamic could account for the significantly elevated prevalence of pendelluft, as observed in our study during SBT with a T piece. The more than twofold increase in pendelluft prevalence during SBT with a T piece in difficult-to-wean patients may also indicate a heightened risk of lung injury in this patient population. It is important to note that, despite the substantial prevalence of pendelluft, the patients included in our study were clinically deemed ready for weaning based on other respiratory parameters and their overall condition, as assessed by the attending intensivist at the time of EIT assessment. From this perspective, monitoring lung ventilation and the pendelluft using EIT during spontaneous T piece breathing could provide valuable additional insights into the actual respiratory function of these patients during the weaning process. EIT may serve as a non-invasive tool to help intensivists and respiratory therapists make more informed decisions regarding the timing of weaning, potentially reducing the risk of complications associated with premature weaning or prolonged mechanical ventilation. Further research and clinical validation are warranted to explore the full clinical utility of EIT in the weaning process.

Our study demonstrated for the first time that the presence of pendelluft independently correlated with increased mortality among difficult-to-wean patients. While Wang et al. previously showed that pendelluft during pressure support ventilation SBT could predict weaning failure [8], their study did not report mortality outcomes. As a retrospective study, we did not observe a significant difference in the occurrence of pendelluft between surviving patients who underwent successful extubation and those who underwent tracheotomy. However, we did detect a significant higher prevalence of pendelluft in non-survivors compared to survivors (93.8% vs. 66.3%). Our previous study found a correlation between the presence of pendelluft and prolonged mechanical ventilation but did not observe a difference in mortality among ICU patients with PaO2/FiO2 ratio below 200 mmHg [14]. The unique focus on difficult-to-wean patients in our current study may account for this new finding. In consistence with our previous findings, the pendellfut group in the difficult-to-wean patient population exhibited significantly longer mechanical ventilation duration and shorter VFDs to day 28 [14]. In addition, the length of ICU and hospital stay tended to be longer in the pendellfut group, providing further evidence of the observed increase in mortality within this subgroup. Though larger-scale, prospective studies are essential for validating our results, the clinical significance of monitoring pendelluft using EIT is underscored by its impact on mortality. Our findings call for future randomized controlled trials aimed at exploring methods to mitigate pendelluft in clinical practice and evaluating the associated benefits.

Pendelluft has been reported to disappear following the administration of muscle paralysis in mechanically ventilated patients with acute respiratory disease syndrome, as indicated in previous studies [22,23,24]. In the case of patients undergoing SBT using a T piece, application of neuromuscular blockers to attenuate diaphragm muscle contractions was infrequent and required close monitoring in clinical practice. However, in certain instances, low dosages of sedatives and analgesics were employed to mitigate excessive breathing effort, alleviate anxiety, and maintain an appropriate level of wakefulness, particularly in cases involving delirious patients [25,26,27]. Some patients included in our study received low-dose continuous fentanyl infusion (15–30 μg/h) during the spontaneous T piece breathing process. Unfortunately, given the retrospective nature of this study, we were unable to obtain the relevant parameters indicative of breathing effort or diaphragm muscle function. However, it is noteworthy that a considerable number of patients (62.5%) who were successfully extubated in our difficult-to-wean population exhibited pendelluft during SBT. A recent study demonstrated that pendelluft occurred during high-flow nasal oxygen therapy could be alleviated by continuous positive airway pressure and further improved with noninvasive ventilation, along with reduced inspiratory effort [17]. In our department, the use of high-flow nasal oxygen as a post-extubation strategy is quite common due to its enhancement in patient comfort. Noninvasive ventilation and continuous positive airway pressure could also be considered. Future investigations may be warranted to thoroughly explore the benefits of these noninvasive post-extubation strategies in the context of pendelluft and its impact on patient prognosis.

We did not find any significant differences in RR, SpO2/FiO2 ratio and ROX index during the SBT between patients with and without pendellfut. Similarly, the PaO2/FiO2 ratio measured not during SBT but on the same day as the SBT showed no statistically significant differences between the two groups. PaCO2 seemed to be slightly decreased in the pendelluft group without reaching statistical significance (39 vs. 40 mmHg, p = 0.057). Previous studies have reported the association between pendelluft and increased EtCO2 and RR [7, 14]. However, it’s important to consider that arterial blood gas analysis was not conducted simultaneously but on the same day as the SBT in our study, making it less reflective of real-time oxygenation and ventilatory efficiency. Therefore, the lack of significant difference in PaO2/FiO2 ratio and PaCO2 between the two groups can be understood. In contrast, the physiological parameters, including RR, SpO2/FiO2 ratio and ROX index were recorded at the time of EIT assessment during the spontaneous T piece breathing process. Consequently, the absence of significant difference in these parameters between patients with and without pendelluft in our study suggested that these traditional global respiratory indicators may not be very sensitive in reflecting lung inhomogeneities during SBT. Regional pendelluft assessment reflects the heterogeneity in lung mechanics and regional intrapleural pressure imbalances, which may offer valuable insights for assessing respiratory conditions.

Our study has several limitations that should be considered. The retrospective design was a major limitation of the study, which limits the ability to establish a causal relationship between pendelluft and mortality. It is crucial to recognize that the association between pendelluft and mortality may be influenced by factors linked to the underlying reasons for difficult weaning, which are beyond the study’s scope. In addition, investigating the correlation between pendelluft and ventilatory drive would be insightful. However, parameters such as airway occlusion pressure or maximum inspiratory pressure are not regularly measured, thus precluding this analysis. Second, the times of EIT examination during SBTs exhibited heterogeneity due to the retrospective nature of this study. Specifically, 83.3% of patients underwent EIT assessments before their 6th SBT. To reduce the potential impact of this variability, only the initial EIT data during SBT was included for analysis. Further study is required to explore the variation of pendelluft in various SBT stages. Third, the causes of death could be multifactorial and extend beyond respiratory failure. The absence of control arm and longitudinal data may not fully capture the comprehensive clinical effects of pendelluft and how it evolves over time. Regrettably, we were unable to obtain EIT data before or after the SBT due to the nature of retrospective analysis, as not every patient included in our study underwent EIT assessment prior to or following the SBT. However, it would be valuable to include EIT data before the SBT to see if there is a difference in the occurrence of pendelluft during SBT and ventilation in future studies. Fourth, the selection of parameters for inclusion in the multivariate binary logistic regression model was mainly based on two categories: demographic characteristics (i.e., age, sex) and clinical characteristics (i.e., pendelluft, SpO2/FiO2 or PaO2/FiO2, mechanical ventilation duration and APACHE II). Because the cases of non-survivors (n = 16) were low, the number of selected variables was limited to those we thought to be important and clinically relevant. Fifth, it may be argued that using pendelluft as a continuous variable might provide more precise results than treating it as a binary outcome. When pendelluft is lower than a certain threshold, it probably represents only noise and not real pendelluft. Therefore, it would introduce error to the regression model if pendelluft was used as a continuous variable. Besides, a much larger sample size is required to effectively utilize pendelluft amplitude as a continuous variable in the multivariate logistic regression analysis. Recently, Wang et al. reported a 3% cut-off to predict the weaning outcome [8]. In our study, we found a 2.5% cutoff value to define the occurrence of pendelluft related to mortality. A search for an optimal threshold for various purposes should be conducted in further studies. In addition, the sample size in our study was relatively small, which could potentially limit the statistical power and increase the risk of type II errors. The broad inclusion criteria and heterogeneity of the included patients may have introduced variability into the data, further affecting the statistical power of the study. Lastly, the restriction of including difficult-to-wean patients, while allowing us to focus on a specific subgroup, may limit the generalizability of our findings to a broader population and may cause bias to the result. However, patients who had no difficulty in weaning were usually successfully extubated after the first SBT and received no EIT measurement. Future prospective studies should consider enrolling this patient population to reduce bias. However, the focus on the difficult-to-wean population in the current study enables us to investigate the association between pendelluft and mortality in the group where this relationship was most likely to be observed.